* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download NBER WORKING PAPER SERIES FINANCIAL OPENNESS UNDER ALTERNATIVE REAL EXCHANGE RATE REGIMES
Survey
Document related concepts
Transcript
NBER WORKING PAPER SERIES REAL VERSUS FINANCIAL OPENNESS UNDER ALTERNATIVE EXCHANGE RATE REGIMES Michael Bruno Working Paper No. 785 NATIONAL BUREAU OF ECONOMIC RESEARCH 1050 Massachusetts Avenue Cambridge MA 02138 October 1981 Prepared for the NBER Conference on "Financial Policies and the World Capital Market: The Problem of Latin American Countries," Mexico City-, March 26—27, 1981. The research reported here is part o the NBER's research program in International Studies. Any opinions expressed are those of the author and not those of the National Bureau of Economic Research. NBER Working Paper #785 October 1981 Real Versus Financial Openness Under Alternative Exchange Rate Regimes ABSTRACT A simple analytical framework is used to consider alternative exchange rate regimes and their bearing on macroeconomic management of a semi— industrial economy. The emphasis is on the implications of different degrees of capital mobility. One of the topics taken up is the conflict between the role of the real exchange rate as a signalling device for long—run resource allocation and the problem of real exchange rate appreciation accompanying the opening up of an economy to short—term capital inflow. Also discussed is the related choice of exchange rate policy as an anti—inflation device. Michael Bruno National Bureau of Economic Research 1050 Massachusetts Avenue Cambridge, Massachusetts 02138 (617) 868—3900 or Michael Bruno Department of Economics Hebrew University Jerusalem, Israel 011—972—2—636015 REAL \'LPUS F UACiAL OPENNESS UD[R I. AL 1[PUA fIVE [YCf'1E RAiL i1 MES Introduction1 The semi-industrialized economy to be discussed here mainly imports raw materials and capital goods and is past the stage of using quantita- tive restrictions for import substitution of consumer goods. Traditionally it depended mainly on primary exports and is now rapidly moving into industrial exports. While it nay still have a multiple protective tariff system and thus be more 'closed' in the real sense than some optimum degree of free trade might dictate, it will rely quite heavily on the use of the formal exchange rate system to handle current account problems. The typical problem for such an economy, unless it happens to be an oil exporter, is policies how to manage its exchange rate and internal macro- over time in face of a large structural deficit, as part of its long-run trade and development strategy. Under reasonable assumptions it will be an optimal strategy for this economy to borrow abroad to finance a large deficit in the early stages of industrialization, use these funds to develop its exports, providing the internal rate of return lies above the marginal social cost of foreign borrowing. independence. Subsequently it will pay back these loans as the economy approaches In the course of the growth process the real effective ex- change rate will usually be required to grow over time, and it should do so at a rate that equals the difference between the internal and external real rate of return to capital (see Bruno 1976). This statement must be modified in the case in which relative technical progress in export - as compared to home - industries (or when the country's relative export price in world markets) is sufficiently high to make actual growth in the real exchange rate unnecessary. Even then it is usually the case that a real exchange rate level has to be maintained over time. -2- Governments, of course, seldom knOw how to sol VC Optimal ty prohl ems, 'or do they know the ny certainty. path of long-run foreign credit availabilities with 1ore often than not, they have to live from hand to mouth. Yet the logic of theoretical constructs may still be followed in at least The marginal cost of foreign loans must not exceed the some respects. internal profitability of the resul ting inclusive structural i nvestment and the real exchange rate must be related to the sue deficit. tar-i ff (or subsidy) of the allowable The main point is that there should be a clear long- run relationship between the ratio of present and future real exchange rates and between the internal and external real rates of interest. simultaneously These affect long—run borrowing, domestic savings and investment policies as well as the real exchange rate as an allocative device. Problems occur when short-term market prices do not give the 'correct' long-run allocation signals for producers or consumers. Two major sources for such discrepancy may arise in the present context, both working in the same direction. The world financial market may temporarily register very low (or even negative) real rates of interest for short and medium run loans2, as it has done for some time in the 1970s and until very recently. The other, often more important, problem is that domestic stabilization policy with its associated restrictive monetary policy, may cause the real marginal cost of domestic credit to rise considerably above the desired long-run rate. Either one of these or.a combined opening up of the gap between the two interest rate signals will, in the absence of other inter- vention, cause substantial short-run capital inflows and a real appreciation of the exchange rate, depressing manufacturing exports and profits, an experience recently shared by a number of countries in the Latin American Continent3 and elsewhere. Prolonged periods of real exchange appreciation accompanying short-run capital inflow, when the long-run horizon dictates the converse, may not be —3- hainful from the point of view of financial portfolio allocations. These are easily reversed at the cost of a phone call. They may, however, be costly in terms of irreversible production and invesijient decisions, which by nature have long gestation periods. This argument, of course, depends on some myopia in the expectation formation process of investors. Given the market inperfections in a typical semi-industrialized country this is a reasonable assumption to make. This type of conflict between short-term and long-term asset pricing also of raises the question of financial openness, as distinct from real openness Real openness relates to the degree of substitutability a country. of goods across borders and the efficiency advantages to be reaped from reducing intervention in the free flow of real trade. Financial openness has to do with the substitutability of foreign and domestic financial assets and the extent of interference with free capital mobility. Since the ex- change rate enters the pricing of both tradeable assets and tradeable goods conflicts of price signalling may occur when financial openness is pursued along with real openness. We shall assume here that the problem of relating long-term borrowing (and investment) to the development strategy has somehow been taken care of and we may, for our purposes, consider the long-run capital imports as largely determined outside the present framework. This leaves the question of openness toward the world financial market for short-run capital movements. Also, how does financial, as distinct from real, openness relate to the operation of the economy under alternativ exchange rate regimes? After laying out a simple analytical framework (Section II) we shall consider the main alternative exchange rate regimes and their bearing on macroeconomic management in the light of the extent of real and financial openness with of the country. The regimes to be considered are a fixed peg, or without a 'crawling' commercial policy, the crawling poj, and the Iil3naged float (Sections I1I—V). market, we shall put the na This being a conference on the capital in eniphasis on the implications of di fferent degrees of capital mobility tinder these regimes. There is a separate question of how openness restricts or enhances brief transition periods it nay pay stabilizing device, even at the last domestic cost extent of real and financial stabilization policy. to use the of 'wrong' section (VI) considers the two—iay adjusnent the exchange rate long-term Thus for as a price signalling. The relationship of exchange—rate and the inflationary process. Should the exchange rate always accommodate to the inflation rate (or overtake it), or can it sometimes be used in the opposite direction tornoderate an excessive rate of inflation? While this and other problems seem to be a hot subject of debate in Latin American scene, I do not pretend to know enough about the empirical background of these countries in order to go into analysis of actual the examples. The discussion has partly been inspired by the experiences of my own country (see also Bruno and Sussrnan, also 1979, 1980) which hopefully may be relevant to a wider set of countries. II. Analytical framework We can represent the general equilibrium of the economy in a diagram in which the real exchange rate (q = e/p) and real money balances (m = M/p) appear on the two axes.4 We shall consider the commodity market, asset supply and demand, and the balance of payments. Coimnod i ties Suppose the economy produces a composite good (Q), priced p. which is an imperfect substitute with a world export good priced abroad at p* = Exports will be a positive function of their relative prices, q(l + tx), where q (= p e/p) is the subsidy. Imports (N) in this economy are assumed to be an imput into the real exchange rate and t is a possible export productive system, together with labour (L), and capital (K), i.e. we have - Q Q(L, N, K). rhe short-run 5upply function is given by where w is the real wye ( W/p), q(l + t)p; QS[J is a tariff rate and p is the relative world price of the import. Domestic absorption depends on real income, the real morley supply (m = M/p), expected inflation ( and a fiscal impact variable (g). Under the usual assumptions we can represent the excess demand for goods in the form: () QdQSf(qm;u) where U1 stands for a vector of various exogenous shift factors (g, w, t, t, -K).6 The downward-sloped curve QQ in Figure 1 gives the combinations of real exchange rate and real balances that keep the comnodity market in equilibIt will shift inwards with an increase in the government rium (Qd = QS)• deficit (g), the real wage (w), expected inflation (r), or any of the other Above the curve there is excess demand and prices factors listed under U1. will tend to increase; below it there is excess supply and prices may or Since the price level may not decrease depending on price stickiness. appears in the denominator of both variables on the axes, the movement of prices in this diagram will always take place along a 450 vector, towards the QQ curve. The_basic_balance The basic balance (T1) will here be defined as the difference between exports and current import requirements plus exogenously given long-run capital flows (F). The reason for distinguishing between current import rLquir1ents and what are normally registered as imports is to separate out element in imports and add it to the capital account. inventory niovements i nvestinent) in . the current will The idea be a form of account is meant to Thus, iiurt a di ticn a bi eg rt of cut art of pi tel arts :5 u t S of aviuys ceparatng ( the Irdng disti FIJI n ish 1et en icrg rUn t:n' 1er and loans which are usually handled and determined centrally by goVernments and will riot affect private sector wealth directly, and short-term capital flows (and changes in exchange reserves) which are the only ones to be considered endogenous here. For simplicity, we assume F to include foreign investment and to be defined netof interest payments on the outstanding debt. We can now write = (2) iihcre 11(q, + U2) + is another vector of exogenous shift factors U2 Current account balance is represented by the 1, at a given real exchange rate q0 = e0/p0. deficit in the basic balance (T1 < (F, t, t) horizontal line in Figure Below the line there is a 0), above it there is a surplus (T > 0). The adjustment when out of balance depends on the total balance-of—payment position and the exchange rate regime of which discussion is deferred for the moment. The NX line shifts up with a fall in F or an overall reduction in import tariffs and export subsidies.8 Money and other assets There are three kinds of privately-held assets in the system in addition to the capital stock: government bonds (B9) carrying a nominal interest rate i, domestic money (M), and a foreign currency denominated asset (Z). The latter could, in principle, include paper money, an interest- bearing asset (net of debt and carrying an interest rate i*), or inventories of imported goods. For simplicity, we shall ignore the foreign interest rate and only take into account the expected rate of devaluation (E). Money demand (Md) is a negative function of the nominal rate of interest (1) and is proportional to gross output (Q). equilibrium with given money supply (11). (3) 11d1 = ()Q = The money market is in -7- rates of For high As expected inflation may be substituted for i in (3). is usually assumed, only drjmestic residents hold M, while both domestic and foreign residents may hold 7. asset (7d) is assumed The relative demand for the foreign to take the following simple form (see 1iles 1978; Calvo and Rodriguez 1971): qzdimd ezd/Md (4) where q more c/p as elastic If k(s) before, and c is the expected rate of depreciation.9 The k, the closer substitutes are M and 7. total short—term foreign exchange assets hold in the economy are B and Central Bank foreign exchange reserves are A, actual private assets (Z) will always equal the difference Z = B — A. B may change only as a result of an imbalance in the basic account as defined above, thus, it will be assumed to be given at any moment of time. Foreign exchange reserves (A) may or may not be determined by the Central Bank depending on the foreign exchange regime (see below). If both the markets for .1 and Z are in balance, equation (4) can be turned into an asset market equilibrium relationship: q = mk(E)/(B-A). To allow for different This is represented by the line ZM in Figure 1. degrees of financial openness we assume that the actual change in the privately—held foreign exchange asset (AZ) is proportional to the desired change - Z0 from any initial level, Z0 (3) equilibrates. =B - A0, while the money market (This assumes that the bond market adjusts more slowly.) AZ = [k(c)m/q - (B0 - A0)1 (5) where 0. The relationship between asset formation and the balance of payments is taken up next. -0balance of ayiiants The The balance the and of payments consists of the sum of the short-r-un capital account, which in of private foreign-exchange assets (-AZ). (1 can = 1 + 12) must equal the basic balance (Ti) turn equals the net sale The overall BoPsurplus net accumulation of Bank reserves (AA). We thus write: (6) 1 = T1(q) + Bwhere A - k(c)m/qJ = T(q, m; U3) = AA U3 includes all the shift variables appearing in U2 as well as (B0 - A0, — The line ZM in Figure 1 represents the short-run capital account balance = 0). Its slope is k(r)/(B0 Above it 12 > 0 and below A0). itT2 < 0. The curve TI which represents the overall balance (T = lie between the lines NX and ZM. 0) must Below it I < 0 and above it I > 0. This two-part graphical representation of the BoP, which was inspired by Frenkel and Rodriguez (1980), is also useful for considering different degrees of capital niobility.1° A very low level of (relatively flat TI curve) would correspond to the case ofrather effective exchange control. In the extreme case ( polar extreme ( = 0), = ) is only the current account matters. The other the one in which only asset behaviour (12 matters. There is instantaneous adjustment of z to 0) (as in Dornbusch 1976). The banking balance sheet We end this preliminary description by stating the basic balancesheet restriction of the banking system, within which monetary policy operates. Denoting Central Bank credit to the government by C9, to the private sector by C, and private banking credit to the private sector by we have: -9-. C9 C + C. cA M (1) subsequent reference we also note that by adding the net foreign For exchange asset eZ to the LHS of (7) and remembering that A + Z = B we get another constraint on the sum of the two assets. M feZ = (8) where C III. Cg 1 C eB+C + C is total bank credit. Ope ingaF We rLimitedatal Mobility first consider the process of large step adjustment of the exchange rate in response to shifts in the basic balance under a pegged exchange rate regime. An important issue in the present context is the limitation imposed by the role of expectations of exchange rate change and of short-run capital movements even when the economy is relatively (though not absolutely) These pose considerable short—term closed to financial capital movements. stabilization problems, especially for monetary policy. Let us discuss these problems and possible remedies. Consider a fixed exchange-rate economy with initial equilibrium at a nominal exchange rate e0, quantity of money M0, and price level p0 (see Figure 2). Cormmodity market equilibrium is given by the curve QQ, the basic balance is given by NX. There is limited capital mobility (low ), so that the balance of payments equilibrium is given by the line TI whose slope is assumed to be between 0 and 1. rate change ( There are given low expectations of exchange- which may be zero). Suppose now that for some reason there is a shift in the basic balance (NX shifts up) which may be due to a long-term fall in lending (F). for. For simplicity account and A real devaluation from q0 to q1 is called assume that the change is confined to the current QQ is not ininediately affected)2 A possible new equilibrium for the economy would be at the point of B which is given at the intersection NX' and the old ZM line, assuming that after the change its slope —10 — - A) and Z rain s.. A set of p'lines ::ight bring thiS about could be a cor:U nation of a no;i n31 doal uaticr to , fiscal the tt and incomes restraint, a reduction in g or w (shifting QQ to QQ'), plus an increase in M to so that Unless prices are do.n— el/CO. ward rigid the same real rovemorit in M/p and c/p could also be achieved by a fall in p. keeping M constant devaluation, ( Me), coupled with a smaller nominal lie ar-c less interested in the exact final outcome but rather in the dynamic sequence of events that may typically occur once the need for a step devaluation is signalled. It is in the nature of a fixed exchange-rate regime that there is a considerable lag between the first signal of a required change in e and the actual act of a devaluation, let alone the implementation of the internal demand restrictions that have to accompany it. if' and The moment tr shifts towards foreign exchange reserves start falling (AA < 0), two processes may be set in riotion, the one having to do with expectations, the other with fluctuations in the money base. As expectations of a devaluation rise (c1 > o' there is import hoarding and delayed repatriation of export proceeds. The asset equilibrium curve rotates upwards to ZM', the balance of payments equilibrium line shifting to TT". say, with The result is a more intensive fall in reserves than would otherwise take place, signalling a larger devaluation than may objectively be required. This can be illustrated as follows: given the expectation of a devaluation, a new temporary equilibrium could take place at the point H. One way of exactly reaching H could be a nominal devaluation from A to G keeping M constant, followed by a price increase (due to an inflationary gap in the goods market) which is represented by an inward move along a 450 vector from G to H. Whether the real exchange rate is above or below the level balancing the new basic account (NX') depends on the extent of the excessive expectations-induced shift of IT" which, in turn, depends on the assetdemand response to expectations change (k/3c) and the effectiveness —1 I of foreign exchange control (E) . Once a icvlu tion hs t ' en expectations will stabilize again (at r, or soue lo';ar ZM returns back to its previous position and BoP sented by TI'. renewed capital At the point H, for example, place, rii:ber). equilibrium there is now Suppose is repre- a surplus with inflow (repatriation of export proceeds, dishoarding of import inventories, etc.) which more than compensates for the remaining basic deficit, and reserves will start rising again. So far we have ignored changes in nominal money which can he assumed away only if there is Central Bank sterilization of the fluctuations in exchange reserves.13 nay This brings us to the sc-cord dynamic process which take place simultaneously with the previously described one. Let us go back to the point A when foreign exchange reserves start falling. Unless sterilized, there will be a fall in M which in Figure 2 would show as a horizontal leftward move from A towards D. Excess supply and unemploy- ment develop unless prices (and wages) are downward flexible. Again, a possible move under price fleibi1ity would be a fall in M by the amount DA and a reduction of prices along a 450 vector from D to E. If expecta- tions of a devaluation are kept at zero, the BoP would have reached temporary equilibrium at E without a devaluation but through a monetary squeeze and a forced deflation. If prices are inflexible, the monetary contraction and the deepening slump could in principle continue until an unemployment equilibrium is reached at F.'4 It is the capital flows in response to the interest rate gap15 that will finance the basic deficit. Obviously, this is neither a sustainable nor a socially desirable equilibrium. What usually happens is that the internal pressure on the monetary authorities makes for monetary expansion to counterbalance the loss of reserves. The counterpart of this phenomenon, after an excessive devaluation, is the reverse upward pressure on the money supply coming from the renewed capital inflow and the rising reserves. Suppose we are temporarily at H -12- when BoP equilibrium is given by TI'. tack of monetary restraint (and legitimdte pressure on the part of sectors hurt by the real credit squeeze) will cause H to rise. In terms of Figure 2 there a horizontal line from H towards iT1. pressure At the same time inflationary develops causing a rise in prices (move back towards QQ). The large swings in short-run capital movements and in the money pre— and post- large devaluations are hard to and will be a movement along contain even under exchange control constitute one of their main drawbacks (see Bruno and Sussman an account of the Israeli experience). 1979, for The pressure to keep real money and credit high (e.g. for investment) may cause a continued inflationary process driving down the real exchange rate to a renewed BoP deficit, unless the devaluation is coupled with the appropriate measures that will shift QQ outwards.16 If exchange controls are tight ( 0) and the original level of real balances is desired,an equilibrium devaluation could be reached at C providing there is a sufficient accompanying budget cut to move the QQ line through that point. If controls are not tight and there are capital flows, temporary equilibrium could be reached at N, say, with a smaller budget cut and smaller real devaluation. This, however, can only be temporary since the current account deficit will gradually erode foreign exchange assets and require further cuts in the budget deficit, and an additional increase in c/p to make room for the expanded production of tradeable goods, as IT' gradually shifts up. Only a larger budget cut to QQ' (and/or larger drop in the real wage, coupled with a rise in m and q) can keep all markets in new current-account equilibrium at the point B. The main problems that arise in the operation of a fixed peg which is adjusted in large jumps thus center around the sizeable accompanying fluctuations in real reserves and in the money supply and the large one-time adjustments requi rd in other expenditure items. The latter are unavoidable wdr my chmge rite regime since a rral rce teisfer 'ill alys be requi red (unless there is a slack in the system which can he diver ted to the tra'ieable g ods sector). arid the exchange real On the other hand, the monetary upheavals loss of reserves could in principle be delayed if the real rate corrections are moving to a regime confined to the current account. Short of of small-step adjustient in the formal rate one nay operate the fixed formal peg on capital account transactions while correcting the real effective exchange rate in smaller steps through adjustments in commercial policy. Consider the case in which we have a pure dual rate --a uniform tariff on exports. (tn) on imports When F falls, on the two accounts and a subsidy of the same rate (t = t) as in the above discussion, one could in principle increase t and t so as to keep the N line in place, while manipulating the the budget so as balance is in deficit). to compensate for the loss in revenue (when In that case the economy may stay in equilibrium at A while at the same time reducing the basic deficit. Countries hardly ever operate a uniform tariff rate on imports. However, a close substitute for the above idea is a flat export subsidy based on value added. Israel had operated such a system quite successfully through the l960s and part of the 1970s. Large devaluations would take place only once every few years and in between the rate of export subsidy (and sometimes also the tariff rate on imports) would be adjusted upwards gradually. problem with such a system is usually twofold. The One is that it tends to get abused once the subsidy rate reaches high levels. There is considerable incentive to cheat in the calculation of value added which leads to trade distortions and the like. The other difficulty is that such measures usually meet with strong disapproval of the It can be claimed that when such a international agencies (GATT, THE). used in small measure and with some prudence system may nonetheless have its merits. shall see, have their drawbacks too. The alternatives, as we -14- IV. The rros aid Cons of a Cra:iing Peg We now keep the assumption of a formal peg but assume that the adjustmerits are made in very small steps and at frequent intervals.11 The advantage of a crawling peg over the previous regime lies in the two areas where the large-step adjustment regime is weakest. It usually avoids the political and social tabu attached to an act of currency depreciation and shifts the exchange rate adjustments from the front pages of the newspaper to back-page financial columns. crises The major exchange reserve and monetary upheavals may be avoided. One by-product of the crawling peg is that no major political decisions or restrictive internal policies have to be taken at each step. This apparent advantage of mini-devaluations may paradoxically sometimes also be a source of major weakness. In a government-deficit and inflation-prone economy the operation of the exchange rate is delegated to a ministerial or Bank committee without the accompanying continuous urge for small-step fiscal and incomes policy discipline.18 The fact that devaluations now take on a more continuous form make for built-in inflationary expectations (i 11 = c = e/e), thus exacerbating inflationary pressures. Under a crawling peg, especially if i.t is predetermined, the capital account curve no longer fluctuates. Suppose expectations are fixed at and the relevant asset demand curve is ZNI' in Figure 2, with equilibrium momentarily at K. required. Say a small step adjustment to a new BoP curve IT" is The small triangle marked by the points K, L, H marks the policy trade—offs. If fiscal policy can be used along with a mini-devaluation we may move from K to L or at least refrain from getting a fall in real balances (i.e. maintain real credit levels). If fiscal or real wage restraint cannot take place, real money balances will have to fall as we move from K to H, by a cut in nominal money, by inflation, or by both. problems One of the of the crawling peg is the continuous built-in pressure for monetary restraint to eep doiestic iriteret rates high reldtive shon by the fact that at the is N or ioint K on LM' , in has to (j* ) . This to be lover than at at B. V. High Capital Mobility and Flexible Rates So far we have considered regimes in which the authorities directly peg the exchange rate in which case the BoP need not be in momentary equi]i brium and foreign exchange reserves are the shortrun buffer. In the Introduction we posed the problem of conflicting longterm and short—term price This may occur in a situation of relatively high capital signalling. mobility and exchange rate flexibility, to which we now turn. Consider the case in which the exchange rate automatically adjusts to equilibrate the exchange market. It helps to discuss first the case of a flexible rate under perfect asset niobility ( = 0) and no intervention (AA = 0). In an inflationary situation and when the foreign asset is highly liquid (k(c) is elastic), it may sometimes be advantageous to use total bank credit (C = Cg ÷ C + C) rather than M as the financial instrument. The commodity market equilibrium curve QQ can be expressed in teniis of c C/p (instead of m) and e/p, keeping all other properties as before (see Figure 3). The asset market curve can be transfonned by substituting from equation (8) into (4) and writing (9) c/p = _() B - = B - A to get: C/p. [1 + k(c)]A Equation (9) is represented by the curve ZC in Figure 3 which has analogous properties to the curve ZM. It is a ray through the origin whose slope increases with c and A and falls with B. Equilibrium takes place at the intersection of the asset market line (ZC) and the com;odity market QQ (point A). In case of a disturbance the exchange rate immediately adjusts to equilibrate the asset market first. Consider, as before, the case in which there is an increase in the basic deficit (NX to NX') and for simplicity assuma first that there is no change in the cc nodity market. The only channel by which this may have an effect on the economy is asset the market, through a possible impact effect on expectations (c) and a effect on the total supply of foreign exchange assets (B). there is a one-time effect on c rotating LC to ZC'. The nominal longer-run Suppose (and real) echarge rate will overshoot to the point F and may quickly be eroded by inflation with real appreciation moving the along the line ZC' . but there is gradually from F to G At the point G the BoP is in short-run equilibrium, a sustained basic deficit. This may disappear over time only as B falls and the ZC curve continues to rotate up;ards. equilibrium will take Final place only when the asset equilibrium line and the cornmodity market line intersect at the new basic balance equilibrium (see point C). The story is only slightly modified if we disturbance that has commodity economy consider that the shifted the basic balance (NX) also shifts the market (e.g. shift of QQ to Q'Q' as would happen in the case of an increase in import prices). There are some clear drawbacks to this automatic adjustment process. One is that in the first stage there may be an excessive with no real benefits to the current account. This is price increase only one example of the 'overshooting' tendency of an exchange regime in which the asset market • takes a major role in the short-run determination of the exchange rate (cf. Dornbusch, 1976, and the related literature). • Similar effects charac- terize the case of unexpected domestic monetary changes. For example, a monetary expansion from A to B raises expectations of a depreciation, causes a jump of the exchange rate to the point F on LC', say, followed by inflation (more from F to G). The case of monetary contraction, from A to H, which may be needed to correct a commodity market imbalance (at QQ') leads an appreciation are downward to (point R) coupled with excessive unemplouent unless prices flexible. The main drawback of asset market determination in the present context, quite apart from overshooting, is that the process of adjustiient to a new long-run situation may be much too slow. signalled by the The real exchange rate that is asset market (at the point G, say) does not long—run rate from the point of view of reflect the true real resource allocation. Given the gestatior lags of exports and of investment in export industries and the imperfections in the expectation formation mechanism this would seem to be the major cost of allowing short-run capital movements to determine the exchange rate. One way of avoiding this signalling problem is to impose a direct tax on capital imports as was done in 1980; Bruno and Sussman 1980). a recent Israeli episode (see Liviatan In Figure 3 this can be shown by noting that the imposition of a tax (T) on capital imports is the same as raising the expected cost of foreign borrowing and thus causing the asset market curve to rotate in the required direction (say, from ZC' to LCD). This may not overcome the overshooting problem but may, at least, avoid over- valuation of the currency.'9 A managed float, which takes the fonu of intervention in the foreign exchange market, in principle works in the same direction. Purchase of foreign exchange by the Central Bank raises foreign exchange reserves (A) and also rotates ZC anti-clockwise. What happens when capital mobility is not perfect and we don't have instantaneous arbitrage? Overshooting will clearly be mitigated. Con- sider again the case in which the current account flows also play a role in the market. Suppose that with the projected increase in the basic deficit BoP equilibrium is represented by the curve TI' (shifted from TT and assuming no immediate change in expectations). rate jumps to D and then adjusts along TI' to G. overshooting (and less inflation) in this case. Under a float the exchange Note that there is less If we now also assume a shift in expectations, the new equilibrium will be at a higher point, G1, -13- Obviously, the is the TI 1 irie the imre iiiportent is the basic balance in the determination of market signals and the closer will on QQ. flatter a float bring us to the long-run signal. Does this analysis imply that it is always advantageous to control capital movements, i.e., limit financial openness for the sake of real openness? One should be careful not to jurip to hasty conclusions. it were possible to drai run an absolute distinction between short-run and capital movements in practice, then such a conclusion justified. tinguished However, we should hear If in mind that the long- may perhaps be way we have dis— between capital movements in terms of exogenously determined long-run funds (F) and endogenous short-term capital is somewhat artificial. Excessive foreign-exchange control could also drive away legitimate longrun capital, which may thus affect the options for the real economy (i.e. shift the NX curve). other. The costs and benefits must be weighed against each In any case, the argument for financial closedness may often be confined to particular short-term situations. VI. Exchange Rate Adjustnent and Inflation: Accommodation versus Mdratfôn The last issue to be discussed briefly is the change rate change and inflation. relationship between ex- This brings up the case in which short- run stabilization may take the lead over long-run objectives in exchange rate pricing. Specifically, should the rate of devaluation always be made to accommodate (or overtake) the inflation rate for the sake of the long-run external balance? Or are there situations in which it pays to moderate exchange rate changes as a short-run stabilization device even at the cost of a real loss of reserves? For simplicity we shall conduct the discussion in terms of steadystate rates of change, p and e, and assume consistency of expectations, i.e. n p and n = e. Cue basic siurt-term reiaticnhip bet:en te ratr of iriflatiun nd depreciation in an open economy can be derived from the coodity labour markets and represents In reduced process. (10) form and the price-wage-price dynamic adjustment this can be written as follows: + ue p The tpath-through' coefficient a incorporates the role of the direct and indirect import coeffici ent as iiell as the impi ied wage-price linkages. a is most likely to be less than one, providing we c'o rot have 100 pem cent indexation. The intercept summarizes the role of excess demand factors in the commodity and labour markets as well as autonomous cost-push factors (world price Equation of raw materials, indirect taxes on consumer goods, etc.).2° (10) appears as the line PP in Figure on the -axis and slope a < 1. 4, with intercept ir Given any rate of depreciation (c) this curve gives the implied inflation rate that will be propagated by the real given its underlying behavioural and institutional parameters. system This inflation rate can in principle be lowered either by a reduction in the extent of linkage (a) In or by fiscal and other stabilization measures, lowering practice it is unlikely that the slope can be changed much (except momentarily under special circumstances), while the intercept of the PP line is more amenable to policy change. Line XX in Figure 4 represents the long -run real depreciation that keeps the current account in balance. The intercept 0 (which may be zero or even negative) reflects the effect of the expected long-run reduction in capital flows net of autonomous time-shifts in the basic balance (due to productivity growth, relative world price changes, etc.).2' A crawl that achieves steady state equilibrium under these assumptions is given at the rate c2, which is consistent with an inflation rate of iT2 (see point A in the diagram). More restrictive internal policies (showing in a lower pp' curve) will allow a lower rate of dereciation, c1, with a correspondingly reduced inflation rate (cf. point C with A). If there is a high degree of capital nobility in th sytern, the numentary equilibrium rate of inflation will rut be given by the line XX, but ho:ever, rather by the combination of e and p that keep the asset market in balance over time (line 7Z in Figure 4). This can be obtained by time differentiation of equation (9), leading to an asset equilibrium rate of real depreciation e — p = , where is a positive function of the rate of change of real credit (c), central hank inte'vcntion in reserves (LA), and the rate of change of expectations (c)2 and depends negatively on the basic balance AB). In Figure 4 the intercept of LZ on the p axis measures -. If ZZ happens to coincide with the line XX (ç is then positive), the system is in full steady state equilibrium at A. XX, however, there may be keeps the asset market in a difference if LZ lies to the left and above between the rate of inflation that balance and that which is propagated by the real economy, at any given announced rate of depreciation (compare the point E with A).23 If left to itself, the fall in T will gradually shift the U curve down towards XX. Now suppose that the ZZ curve is deliberately shifted up toits position in Figure 4 through a credit squeeze, and by allowing reserves to fall (—AA > 0). At the same time the authorities preannounce a lower rate of depreciation (c. < c2). take place at a lower rate of inflation (m < these steps are credible.24 real economy, If the is brought down to at Equilibrium can now the point B), providing stabilization program also affects the while the crawl and the expected inflation rate are further reduced (point 0). At this stage the economy may be ready for renewed growth (raise c, E) and a shift to a new long-run rate of devaluation (c1) at C. The sequence of steps A-B-D-C may be a more feasible one to achieve than an attempt at a direct move from A to C. one uses the exchange rate much more (at D, drastically The main difference is that as a stabilizing device is almost zero), but one pays the cost in terms of short-tern debt (or loss of reserves). This is an eple in ihich short-tena capi tal flows are deliberately used as a stabilizing device at the expense of long- tenii signalling. The main problem in practice is the speed at which these adjustments can be made so as to be credible and the size of the real social cost in terms of transitional production and employment slodown. Footno tes Helpful coriients by Pudiger Dornbusch and other participants at the Conference are gratefully acknowledged. The typical loan in the new post-19/3 2 private capital market, even if taken for a long run investment purpose, is basically short--run in nature, in so far as the interest rate has to be renegotiated at short--run 3 intervals. As noted in Diaz-AlejandrO's paper for this Conference, this problem was also relevant in the 1930s. 4 At a later stage we shall put total real banking credit, instead of m, on the axis - see Section V. enter firm asset market equilibrium (see below). 5 n and iT 6 In this formulation we implicitly assume some i-cal wage (w) rigidity, in which case the labour market need not clear. wage case one should include — assuming full employment. remain the same. consist - 7 In a fully flexible [ as the alternative parameter in U1 The basic form of equation (1) would The negative sign placed on K makes the vector U1 only of positive shift factors. For simplicity we ignore the need to modify (2) hen the commodity market is not balanced. Actually the NX line should have a kink at the point of intersection with QQ. When Qd < QS, imports are a function of Qd which in turn depends positively on in. 8 When we are in a multiple-tariff world, this need no longer be the case. Sometimes a tariff-cutting process may actually improve the current account. Even the case schematically given here, where there may be a common tariff on imports which is different from the export subsidy, could lead to this result. 9 The demand for government bonds (Bg) is not specified separately since it can be derived as a residual once total asset demand (supply) is given. If the latter is denoted by a, we get from (3) and (4) that -23- 9 (Contd.) the deiand for real bonds (bd) is : bd = a 10 We differ - (i) Q[l + k (c) ]. from their presentation in using the pair of real variables (q, in) on the axes rather than the nominal e and p and having T. The capital flow term in their paper separate from the goods market. is written directly as a function of the interest differential Li (j* + c)j. There is some advantage in relating the capital flow to the underlying asset behaviour. ii If one adds B on the LHS we get total financial assets(M + eZ + B9) as financed from the three major sources: the basic balance (eB), the government deficit (Cg + B9), and the banking system finance to the private sector (C + Ce). 12 An exactly analogous situation would be created by an inflationary shift in the commOdity market. Suppose we start at B on QQ' and NX' and for some reason (e.g. an increase in the budget deficit or in the real wage) QQ' shifts inwards to QQ. Assuming the line AB to be on a 450 vector, inflation will move the economy back to equilibrium at A. 13 From here on the analysis is the same. Equilibrium at H or at E here implies a fall in real balances from to M0/p1 through a price rise. 14 Strictly speaking equilibrium will probably take place at a point right of F (since now Q = 15 < QS)• This may sometimes take the form of active government encouragement to firms to take foreign credit which is expensive from a social point of view but is relatively cheap given the prevailing higher domestic interest rates. 16 The analysis was conducted with a given commodity market curve. When QQ shifts inwards due to a rise in inflationary expectations or in real ages these problcms are clearly accentuated. -24- 17 In the rsrael i case the crawl irig peg was operated for two years, between June 1975 and October 1917, and at rules but then became quite flexible in teniis of size of change and allowed 18 g. first follo'ed rather rigid frequency It may of adjustment (see Bruno and Sussrrian, 1930). easier to convince the trade be unions of the need to forego one step in the indexation mechanism in a national crisis situation than 19 to reduce indexation on a conti nijous basis. Note, however, that the impos i tion another rates negative has by—product in the form of high domestic interest which may harm investment. The movmrent from A to C obviously a real credit squeeze which involves in of a tax on capi tal imports may be voluntary, through a fall C, or involuntary, through inflation. Ifwewritep=e+(1—)w+d,wyp+wegetp.+ae 20 where a = if n, = nrd and only if y = 1. + (1 )2j, and r = Li - (1 a - 1 This can also be expanded to incorporate real balance effects. 21 If we denote the autonomous time shift of 11 by of T with respect to q by 22 1A and E will and the elasticity one can show that c + be zero in equilibrium, but appear here since they can be used as policy instruments, along with c. 23 One could argue that in this case the expected rate of inflation may be T3 while the actual rate is 1r2. If is maintained and expecta- tions play an important role, actual inflation may gradually rise above 2 24 (does this accord with developments during a hyperinflation?). Also in line with the previous footnote - inflation may for a time rise above 2 before it eventually comes down to rr. References Bruno, M. (19/6), "The Two-Sector Open Economy arid the Real Exchange Rate", American Economic Review, 66, September, pp. 566-577. Bruno, M., and 7. Sussman (1979), "Exchange-Pate Flexibility, Inflation and Structural Change: Israel under Alternative Regimes", Journal of Development Economics, 6, pp. 83-514. (1 9R0) , "Floating versus Crawling: Israel 1 917—19 by Hindsight", Falk Institute Discussion Paper No. 803, January. Calvo, G.A. and C.A. Rodriguez (19/7), "A Model of Exchange Rate Determination under Currency Substitution and Rational Expectations", Journal of Political Economy, 85, June, pp.611-626. Dornbusch, R. (1976), "Expectations and Exchange Rate Dynamics", Journal of Political Economy, 84, December, pp.1161-1176. Frenkel, J.A. and C.A. Rodriguez (1980), "Exchange Rate Dynamics and the Overshooting Hypothesis", NBER, August. Liviatan, N. (1979), "Anti-Inflationary Monetary Policy and the Capital Import Tax", Working Paper, Department of Economics, The Hebrew University of Jerusalem. Miles, N. (1918), "Currency Substitution, Flexible Exchange Rates and Monetary Independence", American Economic Review, 68, June, pp.428-436. —— q e/p ZM 12 > 0 0 TI NX / QQ 0 m = M/p Figure 1 -27- q LM' (c) 7ti(F0) TT" IF e1/p0 e0/p0 QQI QQ 0 m M0/p0 Figure 2 -28- q e/p IC' zcIt 7C 'TI N X' e1/p0 e0/p0 QQI QQ C = C/p 0 Figure 3 —29- p,IT iz / xx E Pp 0' 0 tO £2 Figure 4 e, £